An Ant Colony Algorithm for Roads Extraction in High Resolution SAR Images

Abstract

This paper presents a method for the detection of roads in high resolution Synthetic Aperture Radar (SAR) images using an Ant Colony Algorithm (ACA). Roads in a high resolution SAR image can be modeled as continuously straight line segments of roadsides that possess width. In our method, line segments which represent the candidate positions for roadsides are first extracted from the image using a line segments extractor, and next the roadsides are accurately detected by grouping those line segments. For this purpose, we develop a method based on an ACA. We combine perceptual grouping factors with it and try to reduce its overall computational cost by a region growing method. In this process, a selected initial seed is grown into a finally grouped segment by the iterated ACA process, which considers segments only in a search region. Finally to detect roadsides as smooth curves, we introduce the photometric constraints in ant colony algorithm as external energy in a modified snake model to extract geometric roadsides model. We applied our method to some parts of TerraSAR-x images that have a resolution of about 1 m. The experimental results show that our method can accurately detect roadsides from high resolution SAR images.